Network Architecture and Incentive Mechanism

6.1 Network Architecture Overview

Our education network is designed to connect real-world teaching activities with Bittensor’s decentralized incentive system through a straightforward and transparent architecture.

The system brings together three main components: the education platform, YouTube’s broadcasting and analytics infrastructure, and the Bittensor subnet. Each serves a distinct purpose while contributing to the operation of the broader ecosystem.

The education platform provides the environment where tutors and students connect. It manages tutor profiles, teaching plans, lecture scheduling, agency affiliations, and the services required to organize live educational sessions.

YouTube serves as the initial broadcasting infrastructure, allowing tutors to deliver live lectures through platform-managed channels while providing measurable audience engagement data.

The Bittensor subnet provides the competitive incentive mechanism. Through the subnet, independent miner agencies contribute teaching activities, while validators evaluate their measurable performance and submit weights that influence the distribution of network rewards.

The overall process follows a simple sequence:

Tutor Affiliation → Live Lecture → Audience Engagement → Analytics Collection → Validator Evaluation → Bittensor Incentives

This architecture allows us to focus on building an effective educational experience while leveraging established broadcasting technology and Bittensor’s existing incentive infrastructure.

The platform is not trying to decentralize every part of education from the outset. It combines centrally managed educational services with decentralized network competition and independently calculated incentive scores.

6.2 Miner Authorization and Tutor Affiliation

Within our subnet, miners participate as independent tutor agencies. Each registered miner hotkey can affiliate with up to five active tutors, including the miner itself if it wishes to teach.

To ensure that teaching activities are correctly associated with legitimate subnet participants, tutor affiliations are established through a cryptographic authorization process.

When a tutor wishes to join an agency, they enter the miner’s hotkey address on the education platform and receive a verification code that expires after one hour. The agency’s miner then uses the subnet’s command-line interface (CLI) to review and authorize the request by signing it with the corresponding coldkey and hotkey.

The platform verifies the signatures, confirms the miner’s registration and ownership information on Bittensor, and checks that the agency has available tutor capacity.

Once the authorization is completed, the tutor becomes officially affiliated with the agency and receives a visible affiliation badge on their profile. Only affiliated tutors are eligible to schedule and deliver lectures through the platform.

Tutors can belong to only one agency at a time, although they may switch agencies according to defined rules. In the initial version, switching is generally permitted after seven days with the current agency, while leaving or being removed involves temporary restrictions on joining another agency.

These rules help maintain stable agency relationships, protect scheduled lectures, and prevent the manipulation of teaching attribution. Lectures that have already started remain attributed to their original agency, while future bookings follow the tutor’s new affiliation after an authorized switch.

Through this system, miner agencies can independently recruit and manage tutors while maintaining verifiable relationships between their network identities and educational contributions.

6.3 Live Broadcasting and Analytics Collection

Our initial platform uses twelve category-specific YouTube channels, owned and managed by the platform, to deliver live educational sessions.

When an eligible tutor schedules a lecture, the platform automatically creates the corresponding YouTube broadcast and assigns dedicated streaming credentials for that session.

Tutors can use supported broadcasting software to transmit their lectures without requiring administrative access to the underlying YouTube channels.

The platform manages the broadcast lifecycle, including scheduled starts, completion, and the cleanup of session-specific streaming resources. A lecture that begins late still ends at its originally scheduled time, and a session that never goes live receives no scoring contribution.

All scheduling and eligibility calculations operate in UTC, ensuring consistent rules regardless of a tutor’s geographical location.

Measuring Live Audience Engagement

The initial incentive mechanism uses YouTube-reported live audience engagement as its primary source of performance data.

Specifically, the platform collects LIVE watch-minutes, representing the estimated total minutes watched during a lecture’s live broadcast.

Viewing activity generated from watching a recording after the live broadcast has ended is excluded from the initial incentive mechanism.

To provide a consistent evaluation period, each completed lecture follows an independent analytics collection schedule:

• 48 hours after completion: The lecture becomes eligible for scoring, and analytics collection begins.

• Every 30 minutes: The platform retrieves the latest available analytics for that lecture.

• 120 hours after completion: The lecture’s scoring eligibility ends, and its analytics record is deleted.

This establishes a 72-hour scoring window for every eligible lecture, beginning 48 hours after its actual completion.

Each lecture follows its own schedule independently. The platform does not wait for other lectures to finish or synchronize all analytics collection into a single global cycle.

Throughout the scoring window, only the latest successfully collected analytics record is retained for each lecture. New results replace previous values rather than being added together, ensuring that repeated updates do not artificially increase a lecture’s contribution.

When collection ends, the platform keeps one number for each lecture permanently: its final LIVE watch-minutes, shown as the lecture’s score on the platform. The detailed analytics record itself is still deleted at the end of the scoring window.

This approach provides a consistent collection framework while keeping data management relatively simple and efficient.

6.4 Incentive Mechanism and Miner Scoring

The initial incentive mechanism evaluates the measurable live audience engagement generated by each miner agency’s teaching activities.

Rather than providing fixed rewards for the number of tutors recruited or lectures delivered, the subnet uses the eligible audience engagement associated with completed lectures to calculate miner performance.

The initial scoring model is intentionally straightforward.

Lecture Contribution = Eligible LIVE Watch-Minutes

Miner Score = Sum of Eligible Lecture Contributions Attributed to the Miner

A miner’s score is therefore determined by the combined eligible live watch-minutes generated by its affiliated tutors’ lectures within their active scoring windows.

Each eligible lecture contributes to the agency to which it was attributed when delivered. If a tutor subsequently joins another agency, previously streamed lectures do not transfer their contributions to the new miner.

How Scores Influence Network Rewards

Validators calculate the relative performance of participating miners using the following reference formula:

Miner Target Weight = Miner Score / Total Scores of All Eligible Miners

For example, consider two miner agencies whose tutors generate the following eligible audience activity:

Miner Agency	Eligible LIVE Watch-Minutes	Target Weight
Agency A	9,000	66.67%
Agency B	4,500	33.33%
Total	13,500	100%

In this example, Agency A generates twice as many eligible live watch-minutes as Agency B and therefore receives twice its relative target weight.

These target weights are submitted through Bittensor’s validation mechanism, where the network’s own rules determine how validator assessments influence actual emissions. Target weights do not represent guaranteed or identical percentages of token payouts.

Importantly, a lecture contributes only during its defined 72-hour scoring window. Once that window expires, its contribution is removed from subsequent calculations, allowing current educational activity to determine the ongoing competition.

The daily, weekly, and monthly rankings shown on the platform add up the same lecture scores for display. They are not a second reward formula.

This creates an incentive for agencies to continuously organize teaching activities, reach interested audiences, and develop educational offerings that encourage meaningful participation.

However, audience engagement is not a direct measurement of teaching quality or learning outcomes. A longer viewing duration does not necessarily mean that students have understood more.

We begin with a measurable, relatively simple indicator, knowing that more advanced evaluation methods will be needed before rewards can reflect demonstrated educational effectiveness.

6.5 Validator Operations and Reward Distribution

Validators play a central role in evaluating miner performance and connecting educational activity with Bittensor’s incentive system.

The platform operator is responsible for collecting and maintaining the latest YouTube analytics, but it does not directly determine the authoritative miner scores or assign validator weights.

Instead, validators independently calculate scores using the published incentive rules and the analytics records provided by the platform.

Independent Scoring

Validators authorized by the platform operator access a dedicated data interface using cryptographically signed requests associated with their validator hotkeys.

At regular intervals, each validator retrieves a complete dataset of lectures currently eligible for scoring.

This dataset includes the relevant lecture information, miner attribution, eligibility periods, and latest available live audience analytics.

Validators independently process these records, calculate tutor and miner contributions, determine relative target weights, and submit network-compatible weights to Bittensor.

Because validators use the platform’s commonly collected analytics rather than independently retrieving all YouTube statistics at different times, the system reduces inconsistencies caused by changes in YouTube’s reported values between separate requests.

Validators using the same dataset, scoring rules, and relevant network state should produce identical scoring results, although their scores may differ when they retrieve valid datasets at different times.

Independent Analytics Verification

For additional accountability, validators receive read-only access to the platform’s YouTube Analytics data.

This allows them to independently check selected lecture records against the original YouTube-reported values.

For example, a validator may compare a lecture’s reported live watch-minutes with the corresponding data available directly from YouTube. If a meaningful discrepancy persists after the platform’s next analytics refresh, it can be flagged for investigation.

These checks are intended to identify inconsistencies or potentially manipulated data. They do not replace the shared analytics records used for scoring and do not constitute cryptographic proof that every reported value is correct.

Handling Periods Without Eligible Rewards

When all miner scores are zero, or a validator cannot obtain a valid scoring dataset, there is no positive educational contribution to distribute weights toward.

In these circumstances, the initial mechanism uses Bittensor’s burn approach, directing the validator’s weight toward the subnet owner’s UID rather than assigning artificial scores to miners.

This ensures that the system does not generate positive miner weights from nonexistent or unavailable educational activity.

Together, these procedures support consistent, independently calculated incentives and provide a practical way to check the underlying audience data.

6.6 Network Integrity and Technical Limitations

A credible incentive system should be clear about both what it measures and what it cannot yet guarantee.

Our initial architecture balances technical practicality, operational efficiency, and checkable performance data.

The use of established YouTube infrastructure allows us to deliver live education without developing an entirely new broadcasting network. At the same time, Bittensor provides an existing framework for open competition and decentralized incentive distribution.

However, this architecture also introduces important dependencies and limitations.

First, the platform remains responsible for certain centralized operations. These include YouTube channel management, analytics collection, tutor affiliation records, and the authorization of validator access. Validators can check reported analytics against YouTube, but this does not eliminate every form of platform-level trust.

Second, live audience engagement is not equivalent to verified educational value. The initial mechanism cannot establish that every viewer is a genuine learner, that all viewing activity reflects meaningful attention, or that a particular lecture has improved someone’s knowledge or abilities.

Third, network rewards remain competitive and variable. A miner’s score influences its relative weight, but actual emissions depend on Bittensor’s network rules, validator behavior, and broader operating conditions. Neither tutors nor agencies are guaranteed a fixed income from their teaching activities.

Finally, the system depends on external services and their applicable policies. Broadcasting capabilities, analytics availability, data access, and permitted incentive-related uses must remain compatible with YouTube’s requirements. The planned compliance review may require adjustments to the mechanism, and successful integration or future approval should not be assumed before it is established.

These limitations do not change the direction of the project. They define the practical boundaries of our initial implementation and give us a sounder basis for responsible development.

As the ecosystem matures, we intend to explore stronger approaches to participation verification, learning assessment, educational quality evaluation, and incentive design, supported by practical testing and real-world evidence.

The larger purpose is to connect the sharing of valuable knowledge with an open, competitive incentive network. A clear, measurable starting point gives us the technical foundation to move toward more meaningful recognition of teaching and learning as the system develops.